Contact centers originate outbound telephone calls for various reasons. In various applications, the calls may be directed to wireline or wireless telephone numbers and are made for the purpose of either playing an announcement or connecting an agent to speak with the called party after they have answered the call. When connecting the agent to speak with the called party after the call is answered, it may be desirous to avoid connecting the agent if an answering machine answers the call. In various contexts, if this occurs, the agent may disposition the call as being answered by an answering machine. Dispositioning the call involves the agent entering information that characterizes the outcome of the call. One way to avoid the agent being connected to an answering machine that has answered the call is to employ automation that detects and determines whether the greeting received after call has been answered has been provided by a “live” person or a voice mail/answering machine. Once it is determined that the call has been answered by a live person, then the agent may be connected to the call. If it is determined that the call has been answered by an answering machine or voice mail service, then the call may be terminated or an announcement may be played. The process of determining whether the greeting is from a live person or an answering machine is frequently known in the industry as “answering machine detection” (“AMD”).
Accurately determining whether a live person or not has answered the call is important, because it increases the agent's efficiency and avoids wasting time. However, accurately determining whether a live person has answered can be difficult to determine in various situations. Typically, algorithms and techniques that detect a live person versus an automated technology are often somewhere in the 90%+ accuracy range. However, this type of accuracy measurement must be further qualified, since there are different ways of measuring accuracy. However it is measured, increasing the accuracy directly correlates to reducing wasted agent time and increasing agent productivity, so there is an obvious motivation to increase the accuracy of answering machine detection.